combustion synthesis/quasi-isostatic pressing of tic –niti...
TRANSCRIPT
Combustion synthesis/quasi-isostatic pressing of TiC0.7–NiTicermets: microstructure and transformation characteristics
E. R. Strutt Æ T. Radetic Æ E. A. Olevsky ÆM. A. Meyers
Received: 20 May 2008 / Accepted: 30 June 2008 / Published online: 27 July 2008
� Springer Science+Business Media, LLC 2008
Abstract TiC0.7–NiTi cermets were produced by com-
bustion synthesis followed by quasi-isostatic consolidation
while the reaction products were still hot and ductile. The
TiC0.7–NiTi cermets were characterized by differential
scanning calorimetry, room temperature transmission
electron microscopy (TEM), and in-situ TEM (temperature
varied during observation). The matrix of the as-synthe-
sized 20NiTi, 40NiTi, and 60NiTi composites contains
both R and B190 martensites at room temperature. No
distinct R-phase morphology could be imaged. In the B190
martensite, [011] Type II twinning, ð11�1Þ Type I twinning
and (001) compound twinning modes were observed as the
lattice invariant shear (LIS) of the R-B190 transformation.
The [011] Type II twinning is often reported as the LIS of
the B2-B190 transformation, but this is the first experi-
mental confirmation of its predicted presence as a qualified
LIS of the R-B190 transformation. The (001) compound
twinning mode is responsible for the fine structure of the
martensite with a wavy morphology. Nanoscale structures
with a thickness of 5 nm were obtained inside the twins.
Twinning was also observed at the interface with carbide
particles, which confirms that some stress relaxation of the
elastic mismatch occurs. At room temperature, the matrix
of the 80NiTi composite had the R-phase structure, which
appeared with a needle-like morphology. Thermal cycling
resulted in the suppression of the R-phase transforma-
tion. This is the opposite of the behavior observed in
un-reinforced NiTi alloys.
Introduction and objectives
Combustion synthesis, also called self-propagating high-
temperature synthesis (SHS), is a method to produce com-
pounds through gas-less exothermic reactions. Although the
principle has been known for over a hundred years, through
the well-known thermite reaction (Alþ Fe2O3 ! Al2O3
þFe) used to join rails, the large number of applications and
varied technologies stems from the pioneering work of
Merzhanov, Borovinskaya, Gordopolov, and others [1–3] at
the Institute for High-Temperature Synthesis. The subject is
reviewed by Munir and Anselmi-Tamburini [4] and by Yi
and Moore [5].
The second objective of this research (first objective
presented in the companion paper [6]) was to investigate
the effect of TiCx inclusions on the martensitic transfor-
mations of NiTi. Near stoichiometric NiTi undergoes a
martensitic transformation from a cubic phase (B2) to
either a rhombohedral phase (R-phase) or a monoclinic
phase (B190). The introduction of TiCx into NiTi creates
internal stresses that should affect the NiTi transformation
path and martensite microstructure. Differential scanning
calorimetry, room temperature transmission electron
microcopy (TEM), and in-situ TEM with observation of the
evolving transformation as temperature was varied in both
E. R. Strutt � M. A. Meyers (&)
Materials Science and Engineering Program, Departments
of Mechanical and Aerospace Engineering
and Nanoengineering, University of California,
San Diego, La Jolla, CA 92093-0411, USA
e-mail: [email protected]
T. Radetic
National Center for Electron Microscopy, Lawrence Berkeley
National Laboratory, University of California, Berkeley,
CA 94720, USA
E. A. Olevsky
Department of Mechanical Engineering, San Diego State
University, San Diego, CA 92182-1323, USA
123
J Mater Sci (2008) 43:5905–5923
DOI 10.1007/s10853-008-2848-y
heating and cooling were used to observe mechanisms of
accommodation.
Experimental techniques
The processing procedure is described in the companion
paper [6]. The cermets consisted of a NiTi (normal com-
position) matrix and TiC0.7 spheroidal particles. Four
cermets with different NiTi contents have been synthe-
sized: 20%, 40%, 60%, and 80% (on a volume basis).
Energy dispersive X-ray analysis
The chemical composition was first studied by microprobe.
However, since the interaction volume of the electron beam
was larger than the spacing between the carbide particles, it
was not possible to get an accurate measurement of the NiTi
matrix chemistry by this technique. Consequently, energy
dispersive X-ray analysis (EDX) was performed within the
transmission electron microscope to determine the chemical
composition of the matrix. A Kevex high angle detector
(HAD) and a Kevex ultra thin window (UTW) detector were
used. A NiTi sample (SE-508) from Nitinol Devices and
Components Inc. was used for calibration. Table 1 lists the
composition of the NiTi standard as determined by the
manufacturer using a Direct Current Plasma Spectrometer.
The uncertainty is within 1%.
The NiTi sample from Nitinol Devices and Components
was stated by the manufacturer to have a composition of
55.7 wt.% Ni, which corresponds to 50.6 at% Ni. The
composition of this NiTi sample was checked by EDX in
the TEM using the HAD detector. The composition was
determined using both the k-factors supplied by the Emi-
spec analytical software and those determined by the
National Center for Electron Microscopy. Both sets of
k-factors provided compositions that matched the manu-
facturer’s reported composition within the 1% experimental
uncertainty. Since the match with the manufacturer’s
reported composition was somewhat closer for the
k-factors provided by Emispec, those values (KNi = 1.489,
KTi = 1.213 for the HAD detector) were used for the
analysis of the NiTi matrix of the composites. The NiTi
sample was not evaluated in the UTW detector, so all data
from the UTW detector were multiplied by normalization
factors. These normalization factors were determined by
comparing the EDX data from the UTW and HAD detec-
tors taken from the same area of the same sample.
Transmission electron microscopy
Transmission electron microscopy (TEM) was performed
at the National Center for Electron Microscopy on a JEOL
200CX microscope. The TEM foils were prepared by
electrodischarge machining the 3 mm diameter samples
used for the DSC measurements into disks with 1 mm
thickness. These were ground down to 100 lm on a
Buehler Minimet polisher, dimpled to \20 lm, and then
ion-milled to electron transparency at liquid nitrogen
temperature. It is important to note that during the foil
preparation, wax was used to attach the samples to the
necessary fixtures. This required heating the samples to
180 �C, which was above the Af temperature. Conse-
quently, the samples were thermally cycled through the
transformation during sample preparation. However, since
the samples were held at liquid nitrogen temperature during
the entire ion-milling process, the martensitic microstruc-
ture in the thin foils was formed in the bulk condition.
The NiTi morphology formed in the bulk condition was
studied by performing TEM at room temperature. Selected
area diffraction patterns, bright field, and dark field images
were acquired to determine the presence of the B2, B190,and R phases and to characterize their morphologies both
near and far from the carbide particles. In-situ heating and
cooling was used to transform the sample in the trans-
mission electron microscope. The sample holder used was
a Gatan 636 double tilt analytical holder with a temperature
range between -170 and +150 �C. The samples were
heated in stages to 110 �C (above the Af temperature) and
then cooled in stages back to room temperature. Liquid
nitrogen was added to cool below room temperature. In-
situ heating and cooling was used to investigate the
transformation path, martensite nucleation and growth, and
the effect of carbide particles on the transformation. The
in-situ experiments were performed in stages to record
selected area diffraction patterns, bright field and dark field
images. The experiments were also recorded on video. The
in-situ experiments were performed with the zone axis
along the [�1�11]B2, so that matrix phase transformations
could be easily detected from the selected area diffraction
pattern.
To aid identification of the different NiTi phases
observed in the electron microscope, diffraction simulations
were performed using the software Desktop Microscopist
Table 1 Chemical composition
of NiTi standard used to
determine matrix chemistry of
TiC–NiTi composites
Reporting level ±1%
Element Weight percent
Ni 55.7
Ti Balance
O 0.02
H \0.0003
Co 0.0273
Fe 0.1800
Cu \0.0050
C 0.0043
5906 J Mater Sci (2008) 43:5905–5923
123
(Macintosh). The parameters for the R-phase are from Goo
and Sinclair [7] and Hara et al. [8]. The parameters for the
B190 phase are from Hehemann and Sandrock [9]. The
lattice parameters, angles, and atom positions used in the
simulations are shown in Tables 2–4. The orientation
relationships between the parent and martensite phases
were also included in the simulations [7, 10]. The simu-
lated selected area diffraction patterns for the B2, R, and
B190 phases were viewed along the [�1�11]B2.
Differential scanning calorimetry
Differential scanning calorimetry (DSC) was used to
determine the temperatures at which the B2 ? B190 and
B2 ? R? B190 transformations in the NiTi occur. The
following temperatures were investigated:
Rs R-phase start temperature
Rf R-phase finish temperature
Ms B190 martensitic start temperature
Mf B190 martensitic finish temperature
As reverse transformation start temperature
Af reverse transformation finish temperature
The onset temperature (or ‘‘start’’ temperature) for the
transformation is defined as the temperature at which the
heat flow first deviates from the baseline upon cooling.
This is because both the formation of the R-phase and the
B190 phase generate heat that can be detected as an exo-
thermic peak on the DSC curve. When no more martensite
is formed upon further cooling, the DSC curve returns to
the baseline. This point is defined as the finish temperature.
It is important to note that start and finish temperatures for
these phase transformations are not the same, because the
elastic strain energy around the martensite resists further
growth unless an additional chemical driving force (i.e.,
cooling) is available. Upon heating, the reverse transfor-
mations appear as endothermic peaks on the DSC curve.
The transformation temperatures were measured with
a TA Instruments 2920 DSC. Cylindrical samples, 3 mm
in diameter and 4 mm in height, were cut by electro-
discharge machining from the 7.5 cm diameter composites.
The samples were held at 150 �C for 10 min and then
cooled at 5 �C/min to -50 �C. After equilibrating at
-50 �C for 10 min, the samples were then heated back to
Table 2 The parameters used for simulation of the B2 NiTi
diffraction patterns
Lattice parameters: a = b = c = 2.998 A
Angles: a = b = c = 90�Space group: 221 Pm3m
Atom positions:
Atom # x y z Element
1 0 0 0 Ti
2 1=21=2
1=2 Ni
Table 3 The parameters used for simulation of the R-phase diffraction
patterns
Lattice parameters: a = b = 7.358 A, c = 5.286 A
Angles: a = b = 90�, c = 120�Space group: 143 P3
Atom positions:
wck x y z Element
a1 0 0 0 Ti
b1 1=32=3 0.083 Ti
c1 2=31=3 0.009 Ti
d3 0.335 -0.005 0.361 Ti
d3 0.677 0.010 0.684 Ti
a1 0 0 0.457 Ni
b1 1=32=3 0:541 Ni
c1 2=31=3 0.396 Ni
d3 0.318 -0.022 0.854 Ni
d3 0.688 0.210 0.160 Ni
R-phase orientation relation with parent B2:
Parallel B2 R-phase Rotation
Plane 111� �
0001ð Þ 0
Direction 211½ � 2110� �
0
The lattice parameters and orientation relationship are from Goo and
Sinclair [7]. The space group is from Hara et al. [8]
Table 4 The parameters used for simulation of the B190 NiTi
diffraction patterns
Lattice parameters: a = 2.883 A, b = 4.117 A, c = 4.623 A
Angles: a = 90�, b = 96.8�, c = 90�Monoclinic
Atom positions:
Atom # x y z Element
1 0 0 0 Ti
2 0 1=2 0.625 Ti
3 1=2 0 1=2 Ni
4 1=21=2 0.125 Ni
B190 orientation relation with parent B2:
Parallel B2 B190 Rotation
Plane 101ð Þ 001ð Þ -6.5�Direction 111
� �110� �
0
The lattice parameters and atom positions are from Hehemann and
Sandrock [9]. The orientation relationship between the B190 and B2 is
from Otsuka et al. [10]
J Mater Sci (2008) 43:5905–5923 5907
123
150 �C at 5 �C/min. The start and finish temperatures were
determined from the intersection of the tangent to curve at
the point of maximum slope with the baseline.
To study the effect of thermal cycling on the transfor-
mation temperatures and the microstructure, samples were
thermally cycled 30 times. This was accomplished by
manually transferring them between an oil bath heated to
+150 �C and a liquid nitrogen/methanol bath held at
-50 �C.
Results and discussion
Chemistry of matrix
Detailed chemical analysis of the matrix of the TiC0.7–NiTi
samples was performed by EDX in the TEM. Figure 1a
shows a NiTi region bounded by two TiC particles for the
TiC0.7–20NiTi. In Fig. 1c, the profile starts in a TiCx par-
ticle, continues through the NiTi matrix, and then passes
Fig. 1 Concentration profile for
the TiC0.7–20NiTi measured by
EDX in the transmission
electron microscope. (a) NiTi
bounded by TiCx particles; (c)
Ni and Ti concentration
particles; (d) background
particles
5908 J Mater Sci (2008) 43:5905–5923
123
through another TiCx particle. In the NiTi matrix, both the
Ni and Ti intensities decrease uniformly with one another
and with the background spectrum (seen more clearly in
Fig. 1d). This reveals that the apparent decline in the Ni
and Ti intensities is the result of differences in the thick-
ness of the specimen. The sample gets thinner along the
measured profile line, so the overall number of counts
measured decreases. The composition of the matrix varies
along the profile line between Ni47.6Ti52.4 and Ni49.0Ti51.0.
These differences in apparent composition may be real or
they may be due to the preferential absorption of the lower
energy X-rays from Ti-Ka. The average composition,
acquired by rastering over several areas of the matrix, is
Ni48.5Ti51.5. Similarly, the other compositions were also
characterized. The average compositions are Ni48.8Ti51.2
for TiC0.7–40NiTi, and Ni51.2Ti48.8 for TiC0.7–60NiTi.
The compositions of the as-synthesized and aged
TiC0.7–80NiTi samples were measured by using the spot
mode in the TEM. Each of the values reported in Table 5 is
the average of three spot measurements. From these limited
number of measurements, the matrix of the aged TiC0.7–
80NiTi sample appears to have a much higher nickel
content than the as synthesized sample. It will be seen later
that this has a significant effect on Rs.
Transmission electron microscopy
TiC0.7–20NiTi
Figure 2 shows a transmission electron micrograph of the
TiC0.7–20NiTi composite taken at room temperature. Since
the sample was thinned to electron transparency at liquid
nitrogen temperature and not subsequently heated above
As, the structure of the NiTi matrix in Fig. 2 has formed in
the bulk condition. This is important, because NiTi is
sensitive to the state of stress [11]. When martensite forms
in the bulk condition, the surrounding material resists the
large strain associated with the change of shape. To reduce
the strain, different martensite variants form a self-
accommodating structure [12]. This self-accommodating
structure is expected to be different if the martensite is
formed in the thin foil condition, because it is subject to
stress relaxation at the foil surfaces.
In Fig. 2a, the matrix contains regions of light contrast
(labeled A) and regions of dark contrast (labeled B). The
light regions are located in the broad central portions of the
matrix, while the dark areas are located in the narrow
corridors between the carbide particles. Figure 2b shows a
selected area diffraction pattern taken from region A of
Fig. 2a. Reflections appear at the 1/3{110}B2 positions in
the 110h i�B2reciprocal lattice direction. Comparison of the
present results with Fig. 2 of Hwang et al. [13] shows that
this pattern is consistent with that of R-phase NiTi.
Figure 3 shows the morphology of the NiTi matrix in
region B of Fig. 2a. In the left-hand region (labeled A),
there is a large martensite variant that consists of a banded
structure. Figure 4 shows the selected area diffraction
pattern, which has been rotated to correct for the lens
rotation in the microscope. This diffraction pattern is
consistent with [011] Type II twinning in B190 martensite
reported first by Knowles and Smith [14] and then con-
firmed by several researchers [15–17].
In the geometry of the parameters that are used to
describe twins, K1 is the twin plane. It is located at the
interface between the two twinned crystals and is a com-
mon lattice plane in both structures. During the twinning
process, the distances and angles in K1 are not changed. K1
also contains the twinning shear direction g1. K2 (K2 before
twinning and K0
2 after twinning) is the other undistorted
plane. g2 (g2 before twinning and g02 after twinning) is the
only direction in K2 that makes the same angle with K1
both before and after twinning.
The elements for [011] Type II twinning in B190 NiTi
were determined by Knowles and Smith [14] to be:
K1 = ð0:72054 1 �1Þ; g1 = [011]; K2 = (011); g2 =
[1 0.63583 –0.63583]; s = 0.28040. It can be seen that g1
and K2 have rational indices, while K1 and g2 have irra-
tional indices.
In Type II twinning, the two regions are related by a 180�rotation about the twinning shear direction g1. The spots in
the diffraction pattern in Fig. 4 can be indexed consistently
if the two regions are related by a 180� rotation about the
[011]B190. The zone axis for one pattern (labeled M) appears
Table 5 Composition of matrix of 7.6 cm diameter TiC0.7–NiTi composites
Ti atomic percent Ni atomic percent
20NiTi sample 51.5 (±0.7) 48.5 (±0.7)
40NiTi sample 51.2 (±1.0) 48.8 (±1.0)
60NiTi sample 48.8 (±0.6) 51.2 (±0.6)
80NiTi sample (data acquired in spot mode) 49.7 (±0.6) 50.3 (±0.6)
80NiTi aged (data acquired in spot mode) 44.9 (±1.5) 55.1 (±1.5)
The corresponding standard deviations are presented in parentheses
J Mater Sci (2008) 43:5905–5923 5909
123
to be along the [�10�1]B190 and the zone axis for the twin
pattern (labeled T) appears to be along the [1�10]B190. Using
the 180� rotation relation, Knowles and Smith [14] calcu-
lated that the angular differences between [�10�1]M and
[1�10]T is only 1.8�. Figure 4 shows the indexed pattern,
where the common reflection is ð11�1Þ.Figure 3b shows a dark field image of region A in
Fig. 3a that was taken with the operative reflection ð002ÞT.
Since a ‘‘twin spot’’ was used to create the image, the
bands oriented with the [1�10T] parallel to the electron beam
direction appear with light contrast.
In TiC0.7–20NiTi, the [011] Type II twins were observed
as the LIS of the R-B190 transformation, not the B2-B190
transformation. Mandangopal and Singh [18] have shown
that [011] Type II twinning satisfies the phenomenological
theory of martensitic transformations for the R-B190
transformation when the rhombohedral angle reaches the
critical value of 86.2�. The observed twin width ratio in
TiC0.7–20NiTi of 3:1 agrees well with the theoretical value
of 2.69:1.
TiC0.7–40NiTi
TEM revealed that the matrix of the TiC0.7–40NiTi was
predominantly B190 NiTi with some islands of R-phase
located in the center of the broad corridors. Figure 5 shows
a selected area diffraction pattern of a region of the matrix
containing only B190. Figure 6 shows a dark-field image
formed by placing the aperture around one cluster of dif-
fraction spots near the 1/2{110}B2. Three distinct B190
martensite morphologies can be identified. Adjacent to the
TiCx interface, in Region A, the martensite has a wavy
character and each of the bands has an irregular thickness.
In Region B, the martensite is lamellar or plate-like. The
bands are narrow, straight and have a relatively uniform
thickness. In Region C, the variants are somewhat smaller
than those in Regions A and B and appear to form a tri-
angular arrangement.
Wavy martensite
Region A in Fig. 6 shows dark field images of wavy B190
martensite. The top image in Fig. 7 was formed from the
(001)M2,T2 reflection (marked by a single arrow in Fig. 9).
The thickness of the bands varied across their length. For
example, measurements of the apparent thickness of a
single M2,T2 band (light-band) varied between 10 and
30 nm. The average thickness of several M2,T2 bands,
measured across a perpendicular line, was 16 nm.
The bands that appear dark in the top image of Fig. 7,
appear light in an image formed with the aperture around
Fig. 2 (a) Bright-field image of
TiC0.7–20NiTi, showing NiTi
surrounded by TiCx particles.
(b) SAD pattern of Region A
in (a)
Fig. 3 (a) Bright-field of NiTi matrix in Region B of Fig. 2a. (b)
Dark-field image of lamellar B190 in matrix of TiC0.7–20NiTi
5910 J Mater Sci (2008) 43:5905–5923
123
the (00�1)M1,T1 reflection (marked by a double arrow in the
SAD). The lower picture of Fig. 7 shows this image. The
thickness of a single M1,T1 band varied between 14 and
50 nm. The average thickness of many M1,T1 bands was
24 nm.
Wavy martensite has previously been reported in thin
foils that have been thermally cycled or held at ambient
temperature for several weeks [19]. Wavy martensite
has only been reported in samples which also showed the
1/3{110}B2 reflections in the 111h iB2 diffraction pattern
[19]. This is consistent with the current work in which the
B2-R-B190 transformation was observed.
Figure 8 shows that each of the bands of wavy mar-
tensite contains an even finer substructure. These nanoscale
structures have a thickness of *5 nm. Selected area dif-
fraction of these wavy bands gives the pattern of reflections
shown in Fig. 9. The pattern is similar to Fig. 2b in the
work of Madangopal and Singh [18] in which observations
Fig. 4 SAD of lamellar B190
with [011] Type II twinning
Fig. 5 SAD pattern of matrix of TiC0.7–40NiTi composite
Fig. 6 Dark-field image from one cluster of 1/2{110}B2 reflections in
TiC0.7–40NiTi
Fig. 7 Wavy martensite in TiC0.7–40NiTi
J Mater Sci (2008) 43:5905–5923 5911
123
of wavy martensite were also noted. By taking SAD from
within one wavy band, they determined that the substruc-
ture of wavy martensite consists of B190 correspondence
variants related by (001)B190 compound twinning. In
Fig. 9 of the present work, taken from many wavy bands,
there are four separate patterns of reflections. The patterns
labeled M1 and T1 are related by mirror symmetry with
respect to the (001)M1,T1 plane. The patterns labeled M2
and T2 are related by mirror symmetry with respect to the
(001)M2,T2 plane. The overall pattern corresponds to two
sets of (001)B190 compound twins.
The twinning elements for (001)B190 compound twin-
ning, as determined Onda et al. [15], are K1 = (001),
g1 = [100], K2 = (100), g1 = [001], and s = 0.23848. All
indices of the twinning elements are rational. The two twin
crystals are related by both mirror symmetry with respect
to the K1 plane and by 180� rotation around g1.
In Figs. 7–9, the beam direction was [�110]M1//[1�10]T1//
[1�10]M2//[�110]T2, which was perpendicular to the normal to
(001)B190. The (001)B190 twins were viewed edge on. It is
apparent that the twin thickness (on the order of 1 nm) is
much finer than in the [011] Type II twinning mode.
Madangopal and Singh [18] showed that (001) com-
pound twinning does satisfy the phenomenological theory
of martensitic transformations for the R-B190 transforma-
tion when the rhombohedral angle reaches the critical angle
of 86.2�.
Lamellar martensite
Figure 10 shows the electron diffraction pattern obtained
from a region containing lamellar martensite. The pattern,
indexed in accordance with Fig. 7b of Xie et al. [17],
consists of two sets of reflections that are related to each
other by mirror symmetry with respect to the (11�1)M,T
plane. This establishes that the alternating platelets are
(11�1) Type I twins.
The elements for (11�1) Type I twinning in B190 NiTi
were determined by Knowles and Smith [14] to be:
K1 = ð11�1Þ; g1 = [0.54043 0.45957 1]; K2 = ((0.24694
0.50611 1); g2 = [21�1]]; s = 0.30961. K1 and g2 have
rational indices. K2 and g1 have irrational indices.
Fig. 8 Internal structure of wavy bands in TiC0.7–40NiTi with
nanoscale bands
Fig. 9 SAD pattern of wavy B190 martensite with (001) compound
twinning. Beam direction [�110]M1//[1�10]T1//[1�10]M2//[�110]T2
Fig. 10 SAD pattern of lamellar B190 with (11�1) Type I twinning.
Beam direction �110½ �M== 1�10½ �T
5912 J Mater Sci (2008) 43:5905–5923
123
Figure 11 shows dark-field images formed by placing the
aperture around the (110)M and (002)T. The beam direction
was �110½ �M== 1�10½ �T, which was perpendicular to the normal
to (11�1)M,T. The (11�1)M,T twins were divided into two
regions by a ‘cross’ variant. The twin width on the left side
was noticeably smaller than on the right side. On the left, the
average thickness of the M bands, which appear bright in
Fig. 11, was 6 nm. The average thickness of the T bands,
which appear dark in Fig. 11, was 10 nm. The twin width
ratio was 1.7:1. On the right, the average thickness of the M
bands was 9 nm, the average thickness of the T bands was
33 nm, and the twin width ratio was 3.8:1.
In TiC0.7–40NiTi, the (11�1) Type 1 twins were observed
as the LIS of the R-B190 transformation. The phenome-
nological theory for this twinning mode in the R-B190
transformation has not yet been carried out.
Localized twinning
Figure 12a shows the B190 martensite near the carbide
interface. Striations appear within the (11�1) twins. Close
examination of the SAD pattern in Fig. 10 shows a small
spot located near the double arrow. Fig. 12b shows a dark
field image taken with the small reflection near the
(11�1ÞM.T spot. This image confirms that the striations are
localized to the region near the carbide interface and do not
extend throughout the B190 variant. This localized twin-
ning at the interface between the matrix and reinforcement
may be due to relaxation of the elastic mismatch that
results during sample processing [20].
TiC0.7–60NiTi
Figure 13 shows a global view of the area investigated in
the 60NiTi sample. Region A was located near a hole in the
foil that was probably formed during the TEM sample
preparation. The matrix was composed of many variants of
B190 martensite, most of which had a lamellar morphology
as presented for the TiC0.7–40NiTi. The diffraction pattern
for this region, shown in Fig. 14a, was equivalent to the
superposition of Figs. 4 and 10. It consisted of three sets of
reflections, with one set being common to both the (11�1)
Type I twinning pattern and the [011] Type II pattern. The
beam direction in the TEM corresponded to [�110�M,T1 for
the (11�1) Type I twins and to [101]T2//[�110]M for the [011]
Type II twins, where the subscript M is used to index the
set of reflections common to both observed twinning
modes. A series of dark field images, formed by placing the
aperture around the (001)T1, (00�1ÞM, (110)T1, and (11�2)M
reflections, was used to identify the (11�1) Type I twins. The
[011] Type II twins were imaged with the (1�1�1)T2 and
(111)M spots. Figure 14b shows the dark field image
formed from the (00�2)T1 and (1�1�1)T2 reflections.
Region B in Fig. 13 was an area of matrix located adjacent
to a TiCx particle. Selected area diffraction, performed with
a beam direction of [�11�1]B2, clearly showed reflections at
1/3{110}B2 in the 110h i�B2 reciprocal lattice direction. This
indicated that the matrix had the R-phase NiTi structure.
Dark-field images formed from the 1/3{110}B2 spots did not
provide clear pictures of the variant structure.
Region C in Fig. 13 was located in a corridor between
two carbide particles. At room temperature the matrix had
the R-phase NiTi structure.
Region D was located near the junction of two carbide
particles. Figure 15 shows the B190 morphology both near
Fig. 11 Lamellar B190 with (11�1) Type I twinning
Fig. 12 (a) Localized twinning at TiC interface. (b) Dark field image
J Mater Sci (2008) 43:5905–5923 5913
123
and far from the TiCx interface. The diffraction pattern for
this region (shown in Fig. 16a was composed of at least four
sets of reflections. Figure 16b and c shows the indexing.
Both [011] Type II twins and (11�1) Type I twins were
present. The reflections associated with the [011] Type II
twinning mode had the most prominent streaking. These
twins were located in the long lamellar variant furthest from
the carbide. The variants nearer to the carbide were internally
twinned by (11�1) Type I twinning. In most cases the twins
extended outward, away from the carbide. Figure 17 shows
the variant arrangement near the particle junction.
TiC0.7–80NiTi
As-synthesized TiC0.7–80NiTI
Room temperature TEM of the matrix of the as-synthesized
TiC0.7–80NiTi composite only detected the R-phase
structure. Figure 18a shows the morphology. The R-phase
variants formed plates, which appeared as alternating dark
and light blades. The R-phase variants tended to extend
outward, away the carbide particles. Both straight and
sawtooth boundaries are present. Figure 18b–d shows a
series of dark-field images taken by placing the aperture
around the superlattice reflections labeled in Fig. 18a.
Aged TiC0.7–80NiTi
The aged TiC0.7–80NiTi sample was heated to 350 �C.
Since it had a different thermal history than the as-syn-
thesized sample, the microstructures of the samples were
compared. As shown in Fig. 19, no differences were visi-
bly evident. The matrix appeared to be composed of plate-
like variants that extended outward, away from the carbide.
Figure 19 shows the selected area diffraction pattern.
Similar to the as-synthesized SAD pattern, the diffraction
pattern for the aged samples contains large bright spots
with sixfold rotational symmetry and spots at the 1/3
{110}B2. These features are characteristic of the R-phase
Fig. 13 Global view of area investigated in TiC0.7–60NiTi sample
Fig. 14 (a) SAD pattern of
(11�1) Type I twinning and the
[011] Type II twinning. One set
of reflections is common to both
the (11�1) Type I twinning
pattern and the [011] Type II
pattern. (b) Regions of (11�1)
Type I twinning and [011] Type
II twinning in the matrix of the
60NiTi composite
Fig. 15 B190 morphology near and far from the TiCx interface of the
TiC0.7–60NiTi composite in Region D in Fig. 13
5914 J Mater Sci (2008) 43:5905–5923
123
structure. However, the SAD taken of the aged TiC0.7–
80NiTi composite also contains faint spots clustered
around the {110}B2. Comparison of this diffraction pattern
with that shown in Fig. 4b of Stroz [21] reveals that these
spots are from Ni4Ti3 precipitates.
Differential scanning calorimetry results
Figure 20 shows the calorimetry spectra for the TiC0.7–20,
40, 60, and 80NiTi samples during the first and thirtieth
thermal cycle. During the first thermal cycle for the TiC0.7–
20NiTi and TiC0.7–40NiTi samples, only one peak appears
upon cooling from 150 to -50 �C, even though room
temperature TEM showed that the R-phase and B190
martensite coexist. The absence of multiple peaks on these
DSC curves most likely indicates that the R-phase and B190
transformation temperature ranges overlap. Since there was
never any parent B2 phase coexisting with the B190 mar-
tensite in room temperature TEM and the R phase
transformation always preceded the B190 transformation
during in-situ TEM (to be phase discussed below) it was
assumed that the onset of the peak during cooling marked
the Rs temperature. Rs was measured as the temperature
where the tangent to the DSC curve at the point of maxi-
mum slope intersects the baseline. Table 6 lists the results.
Figure 21 compares the Rs temperatures measured in the
TiC0.7–NiTi composites with measurements made by
Miyazaki and coworkers [22] and Gyobu and coworkers
[23]. The results are plotted as a function of Ni content,
because the transformation temperatures in unreinforced
NiTi have been shown to be dependent on composition.
Small additions of Ti cause relatively minor variations in
Rs (and Ms), while excess Ni significantly depresses the
transformations.
The area under the DSC curve is equal to the net enthalpy
of transformation DhP�Mch . For the B2-R transformation, Goo
and Sinclair [7] determined that DhB2�Rch ¼ �3:75 J/g NiTi
Fig. 16 (a) SAD of [011] Type II twins and (11�1) Type I twins near
Region D in Fig. 13. (b) Indexing of (11�1) Type I twins (c) Indexing
of [011] Type II twins
Fig. 17 Variant arrangement near junction of two carbide particles
J Mater Sci (2008) 43:5905–5923 5915
123
and for the R-B190 transformation they determinedDhR�B190
ch ¼�25:5 J/g NiTi. The area under the experimental curves
during the first thermal cycle for TiC0.7–20NiTi and TiC0.7–
40NiTi was -12.6 J/g NiTi and -25.2 J/g NiTi, respec-
tively. The net enthalpy released during cooling of the
composites was always greater than the total energy that
would have been released if the entire matrix were to have
transformed from B2 to R. The extra energy can only be
attributed to the formation of B190. The Ms was defined so
that the area under the curve between Rs and Ms was equal to
-3.75 J/g NiTi. It was assumed that the temperature where
the peak returns to the baseline defined Mf. Table 6 shows
the transformation temperatures. Figure 22 compares the Ms
temperatures measured in the TiC0.7–NiTi composites with
measurements made by Miyazaki and coworkers [22],
Honma [24], and Hanlon and coworkers [25]. Mari and
Dunand [20] measured a Ms of 70 �C in NiTi, TiC-90NiTi,
and TiC–80NiTi composites with a matrix composition of
48.6 at% Ni.
During heating of the TiC0.7–20NiTi sample there were
multiple peaks. Besides the prominent peak between 83
and 104 �C, there are two smaller peaks, one between 6
and 20 �C, and another starting at 62 �C. During in-situ
TEM it was observed during heating that the regions with
retained R-phase transformed to B2 first and then B190
regions transformed directly into B2 without going through
the intermediate R-phase. Since the forward and reverse
R-phase transformations have a small hysteresis [26] the
peak starting at 62� can be attributed to the transformation
of any remaining R-phase back to B2. The large peak
defines As and Af.
Fig. 18 (a) Bright-field image of the matrix of the as-synthesized
TiC0.7–80NiTi. The R-phase variants formed plates, which appear as
alternating dark and light blades. (b) Dark field image of Variant 1.
(c) Dark field of Variant 2. (d) Dark field of Variant 3
Fig. 19 Dark field image of the matrix of the aged TiC0.7–80NiTi
sample. Temperature = -150 �C
b
5916 J Mater Sci (2008) 43:5905–5923
123
The DSC curves for TiC0.7–60NiTi (shown in Fig. 20c)
have a more complex shape than the DSC curves for the
TiC0.7–20NiTi and TiC0.7–40NiTi composites. Specifi-
cally, several additional peaks appear at temperatures
below the main peak and the overall width of the curves is
broader. Multiple peaks on the cooling DSC curve may
occur if the NiTi matrix undergoes more than one phase
transformation (B2–R–B190). They may also occur when
the matrix undergoes a single transformation, if different
regions of the material transform over different tempera-
ture intervals. The area under the cooling DSC curve was
-24.0 J/g NiTi.
Figure 20d shows the DSC curves for the as-synthesized
80NiTi sample. The main peak during cooling begins at a
significantly lower temperature than the peaks observed in
the other composites. Since the room temperature TEM
detected only the presence of the R-phase, there does not
appear to be significant overlap of the B2-R and R-B190
transformation ranges. This suggests that the Ms temperature
of the R-B190 transformation must be below the lowest
temperature reached in the DSC. Previous studies by Mari
and Dunand [20] have shown that Ms is independent of
particle content in TiC–NiTi composites with low volume
fractions of reinforcement. In such cases, the Ms only
depended on the chemical composition of the alloy. This
would suggest that the nickel content in the matrix of the
TiC0.7–80NiTi composite was greater than 51.0 at% Ni [25].
In the TiC0.7–20, 40, and 80NiTi composites the
numerical values of Rs are quite close to those reported in
thin film NiTi and the numerical values of Ms are quite
close to those reported in solution-treated NiTi. In the
TiC0.7–60NiTi composite, there is a 50 �C shift in Rs and a
130 �C shift in Ms. The chemical content of the matrix was
51.2 ± 0.6 at% Ni, so the expected Ms in unreinforced
solution treated NiTi would be between -90 and -40 �C.
Figure 23 shows the DSC curves for the aged TiC0.7–
80NiTi sample. When compared to the as-synthesized
-0.01
0.01
0.03
0.05
0.07
0.09
0.11
0.13
0.15
0.17
0.19
Temperature (Celsius)
Cycle 1
Cycle 30
20NiTi40NiTi
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
-20 0 20 40 60 80 100 120 140
-20 0 20 40 60 80 100 120 140-20 0 20 40 60 80 100 120 140
-20 0 20 40 60 80 100 120 140
Hea
t F
low
(ar
bitr
ary
units
)
Hea
t F
low
(ar
bitr
ary
units
)
Cycle 1
Cycle 30
-0.05
Temperature (Celsius)
Hea
t F
low
(ar
bitr
ary
units
)
Temperature (Celsius)
Hea
t F
low
(ar
bitr
ary
units
)
Temperature (Celsius)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14 60NiTi
Cycle 1
Cycle 30
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
80NiTi
Cycle 1Cycle 30
(a)
(c)(d)
(b)
Fig. 20 Calorimetry spectra for (a) TiC0.7–20NiTi, (b) TiC0.7–
40NiTi, (c) TiC0.7–60NiTi, and (d) TiC0.7–80NiTi samples during
the 1st and 30th thermal cycle (Scanning rate: 5 �C/min). Curves have
been shifted along the vertical axis in order to facilitate comparison of
transformation temperatures
Table 6 The transformation temperatures measured in the DSC
20NiTi 40NiTi 60NiTi 80NiTi 80NiTi (aged)
Rs 67 67 69 16 43
Rf 26
Ms (estimated) 63 64 59
Mf 52 51 25
J Mater Sci (2008) 43:5905–5923 5917
123
TiC0.7–80NiTi DSC data, it is evident that the forward
transformation occurs over a higher temperature interval
and the reverse transformation is composed of three sepa-
rate events. Based on the SAD image shown in Fig. 19, it is
apparent that precipitation of Ni4Ti3 has occurred in the
aged sample. Formation of nickel rich precipitates increa-
ses the martensitic transformation temperatures by shifting
the nickel content of the matrix toward the stoichiometric
composition. It is worth noting that in the aged TiC0.7–
80NiTi sample the expected increase in the titanium con-
tent of the NiTi matrix was not reflected in the EDX
measurements reported in Table 5. Although the compo-
sitional data were acquired in the spot mode, it was not
possible to determine if the probe was focused on the NiTi
matrix or on the Ni4Ti3 precipitates, because the precipi-
tates were not visible in the image.
Although aging was shown to cause a 27 �C shift in Rs in
the aged TiC0.7–80NiTi composite, it was not clear if pre-
cipitation of Ni4Ti3 occurred in the TiC0.7–60NiTi
composite. The sample was never heated above 180 �C and
no evidence of Ni4Ti3 was observed. Lenticular precipitates
were not seen in the TEM and the characteristic reflections
(such as those shown in Fig. 19) were not observed.
Figure 20a–d shows the calorimetry spectra for the
TiC0.7–NiTi samples after 30 cycles. All the transformation
temperatures are lower in the cycled composites. Decrease
of Ms and Mf with thermal cycling has also been observed
in unreinforced NiTi alloys [20, 22]. In those cases, it was
attributed to the introduction and rearrangement of dislo-
cations during the repeated motion of the parent–martensite
interface [22]. Figure 24 shows dislocations introduced by
thermal cycling in the TiC0.7–40NiTi composite.
TEM of the thermally cycled TiC0.7–40NiTi composite
showed that cycling also resulted in a change of transfor-
mation path. As shown in Fig. 25, after 30 cycles, B2 and
B190 martensite were observed to coexist in the matrix of
the 40NiTi sample. The B2–B190 transformation occurred
without the appearance of the R-phase. This is different
from the behavior observed in unreinforced, solution
treated NiTi where thermal cycling tends to favor the
B2 ? R ? B190 transformation. Miyazaki and coworkers
[22] explained the enhancement of the R-phase upon
cycling unreinforeced NiTi by noticing that Rs remains
constant with increasing number of cycles, while Ms
-150
-100
-50
0
50
100
48 49 50 51 52
Atomic % Ni
M-s
Tem
pera
ture
(C
elsi
us) Matrix of TiC(0.7)-NiTi
Composites
Miyazaki, Solution-treated
Miyazaki, Cold work followed by 400 C
Honma, quenched 650~860
Hanlon (quenched from900 C)
Fig. 22 Comparison of Ms measurements
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
-20 0 20 40 60 80 100 120 140
Aged 80NiTi
Hea
t Flo
w (
arbi
trar
y un
its)
Temperature (Celsius)
Fig. 23 Calorimetry spectra for aged TiC0.7–80NiTi sample. Curves
have been shifted along the vertical axis in order to facilitate
comparison of transformation temperatures
0
10
20
30
40
50
60
70
80
48 49 50 51 52
Atomic % Ni
Matrix of TiC(0.7)-NiTi composites
Miyazaki, solution treated, 1000 C, 1 hr
Miyazaki, aged 400C, 1 hr
Miyazaki, cold worked followed by 400Canneal, 1 hr
Miyazaki pre-strained 4.6%, 11.4%, and 22.8%
Gyobu, thin-filmR
s Tem
pera
ture
(C
elsi
us)
Fig. 21 Comparison of Rs
measurements
5918 J Mater Sci (2008) 43:5905–5923
123
decreases. This leads to a broadening of the R-phase sta-
bility range. In the TiC0.7–40NiTi composite, reduction of
Ms with thermal cycling was observed, but the R-phase was
inhibited. Mari and Dunand [20] also observed suppression
of the R-phase in TiC-90NiTi and TiC–80NiTi after 100
thermal cycles.
Figure 26 shows the microstructure of the matrix of the
thermally cycled TiC0.7–40NiTi sample. It shows banded
features similar to those in Fig. 1a of Madangopal and Singh
[18]. Madangopal and Singh [18] argued that the banded
features correspond to ‘‘ghost structures’’ from internal [011]
Type II twins. They suggested that transformation cycling
causes the accumulation and arrangement of defect struc-
tures, which become imprinted in the parent B2 phase.
In-situ TEM
In-situ heating and cooling was used to investigate the
transformation path, martensite nucleation and growth, and
the effect of carbide particles on the movement of NiTi
transformation fronts. The samples were first heated to
110 �C (above the Af temperature). This process was per-
formed in stages to record selected area diffraction patterns,
bright field images, and dark field images at specific tem-
peratures. The experiments were also recorded on video.
Then the samples were cooled in stages to room temperature
in order to observe the martensitic forward temperature.
Again selected area diffraction patterns, bright field images,
and dark field images at specific temperatures were recorded.
In most cases the R ? B19 transformation was incomplete
at room temperature, so liquid nitrogen was added to cool the
specimen even further.
TiC0.7–20NiTi
On heating the thin foil, the regions with retained R-phase
were the first to transform (Region A of Fig. 2). The R–B2
transformation started at 56 �C. When the SAD pattern was
observed at 60 �C, the intensity of the R-phase spots
(located at the 1=3f110gB2 positions) was diminished. At
70 �C, the R-phase reflections were no longer visible. At
71 �C, the B190 variants located at the B2–B190 interface
(formerly the R–B190 interface) began to transform. They
reverted directly to the parent B2 phase without passing
through the intermediate R-phase.
During cooling, the B2 ? R transformation occurred
first. The nucleation and growth of the R-phase transfor-
mation was not directly imaged in the TEM. This
transformation was only identified by the presence of spots
at the 1/3{110}B2 positions in the SAD. The R-phase
reflections were first noticed at 40 �C and appeared stron-
gest in regions that were adjacent to carbide particles. By
29 �C the entire matrix had the R-phase structure.
Fig. 26 Matrix of thermally cycled TiC0.7–40NiTi. Banded regions
are similar to those reported by Mandangopal and Singh [18] and may
be the vestiges of prior martensite twins
Fig. 24 Dislocations in matrix of thermally cycled TiC0.7–40NiTi
Fig. 25 (a) Structure of matrix of TiC0.7–40NiTi sample after 30
cycles. (b) SAD of Region A with multivariant B190. (c) SAD of
Region B with parent B2 structure. Zone axis [001]B2
J Mater Sci (2008) 43:5905–5923 5919
123
With further cooling, the R ? B19 transformation began.
Fortunately, it was possible to record this transformation.
The sequence is shown in Fig. 27a–d. The specimen was
tilted so that the R-phase appeared with light contrast and the
B190 appeared with dark contrast. The first B190 variants
were detected at -12 �C. As shown in Fig. 28, the B190
nucleated at the intersection of a grain boundary in NiTi with
the TiC0.7 particle. In this region, the activation energy for
transformation is lower. As the temperature was lowered, it
grew in bursts, separated by as much as 30 �C intervals. The
transformation front moved along the carbide particle
interfaces and the NiTi grain boundaries before spreading
throughout the matrix. In Fig. 27c the remaining R-phase
appeared as an island surrounded by B190. At -112 �C only
the smallest amount of R-phase remained.
Figure 29 shows the same area of the matrix as in
Fig.3b. The large variant with [011] Type II twinning that
was located on the left-side in Fig. 3b has been replaced by
many smaller variants. This change in variant structure was
not surprising, because the morphology in Fig. 3b was
formed in the bulk condition, while the variant arrange-
ment in Fig. 29 was formed by heating the thin foil above
Af and then cooling it below Mf. The stress conditions in
the two cases were not the same. The B190 that formed in
the bulk condition was subject to tridimensional constraint,
while the B190 that formed in the foil was subject to stress
relaxation at the foil surfaces.
Upon heating, for a second time, the retained R-phase
regions of the matrix were the first ones to transform to the
B2 phase. Then, as the temperature was increased, the last
B190 martensite plates that formed during cooling were the
first to revert back to the parent B2 structure. The B190
regions transformed directly to the B2 phase, without
passing through the R-phase.
Fig. 27 In-situ experiments in 20NiTi. Top four pictures depict the
R ? B19 transformation upon cooling. Lower four pictures depict
the B190 ? B2 transformation upon heating. Starting in top left and
continuing across and then down the temperatures are as follows: (a)
T = -45 �C; (b) T = -65 �C; (c) T = -86 �C; (d) T = -112 �C;
(e) T = 74 �C, (f) T = 80 �C; (g) T = 85 �C; and (h) T = 86 �C
Fig. 28 Nucleation of B190 at the intersection of a TiC0.7 interface
and a NiTi grain boundary in TiC0.7–20NiTi
Fig. 29 B190 martensite formed in thin foil (TiC0.7–20NiTi). Same
area as in Fig. 3b
5920 J Mater Sci (2008) 43:5905–5923
123
TiC0.7–40NiTi
The in-situ transformation of the matrix of the 40NiTi
sample followed the same course as it did in the 20NiTi,
except the numerical values of the transformation tem-
peratures were higher. Again, the formation of the R-phase
was only apparent from the SAD. No distinct morphology
was observed. The R-phase reflections first appeared at
42 �C .
The beginning of the B190 transformation started at
34 �C, when a dark patch appeared in the narrow corridor in
the upper right-hand corner of the image. No further growth
or new nucleation was observed until the temperature was
reduced to 11 �C. At that point, the transformation was
characterized by the sudden appearance of several dark
branches emanating from the same corridor that had pre-
viously transformed at 34 �C. The B190 looked as if it were
a river spilling onto a flood plain. One branch extended
completely across the broad region of matrix to the narrow
corridor in the lower left-hand corner of the image. The next
series of transformation bursts occurred when the temper-
ature dropped below 4 �C. The long branches began to
widen in a series of small jumps and the matrix in the upper
left-hand corner transformed. The transformation appeared
to be complete at -13 �C. However, when other regions of
the foil were imaged and SAD patterns were recorded, it
became evident that the R-phase still existed in the thinnest
areas.
Figure 30 shows images taken at three temperatures:
room temperature, above Af, and below Mf. The wavy
martensite in the upper left-hand corner of the image
maintained the B190 structure even when the sample was
heated to 110 �C.
TiC0.7–60NiTi
Region A (in Fig. 13)
In-situ heating of the foil was performed in a series of
steps, so that the progression of the reverse transformation
could be studied. In Region A of Fig. 13 there was no
noticeable change in the SAD pattern at 50 �C. At 70 �C,
the dark-field image from the (00�2)T1 and (1�1�1)T2 showed
that matrix was partially transformed to B2. At 108 �C,
only a small amount of B190 remained near a carbide
particle.
To observe the propagation of martensite during cool-
ing, the microscope was originally focused on the retained
B190. Dark-field imaging was used, so that any further
extension of the B190 would show up as light contrast
against a black background. No change was observed from
108 to 22 �C.
Fig. 30 TiC0.7–40NiTi imaged in in-situ heating experiment; Top:
T = 22 �C. B190 martensite formed in bulk condition; Middle:
T = 110 �C. B2 NiTi; Bottom: T = 27 �C. B190 formed in thin foil
J Mater Sci (2008) 43:5905–5923 5921
123
Nearby, at 22 �C, SAD showed that the R-phase trans-
formation had occurred, but dark-field images formed from
the reflections located at the 1=3f110gB2 positions did not
show a distinct variant structure. Liquid nitrogen was
added to drop the temperature below room temperature. By
-100 �C, the B190 transformation had begun and the front
was moving away from the carbide particles and toward the
hole. This B190 variant structure, formed in the thin foil,
was very different from that formed in the bulk condition.
Instead of the lamellar morphology shown in Fig. 14, the
B190 variants formed in the foil were smaller and lens
shaped.
Region B (in Fig. 13)
After heating the TEM foil to 50 �C, selected area dif-
fraction was performed on Region B of Fig. 13. The
multivariant R-phase structure that was present at room
temperature had mostly transformed to B2. Only the
reflections from one R-phase variant could be detected, and
the intensity was very low. The transformation to B2 was
complete by 70 �C. During cooling from 108 �C, no
change was detected until 22 �C. Diffraction patterns at
this temperature showed that the R-phase had reappeared,
even though the R-phase transformation front had not been
noticed. At -22 �C, the B190 martensite still had not
nucleated. By -104 �C, the B190 martensite had nucleated
at the carbide interface and was growing outward, away
from the particle. Figure 31 shows a dark field image of the
B190. The B190 martensite variants were lens shaped and
approximately 250 nm long and 50 nm wide. Images
showed that islands of R-phase NiTi still existed in the
broad corridors. SAD confirmed the coexistence of B190
and R-phase NiTi.
Region C (in Fig. 13)
During in-situ heating, Region C of Fig. 13 had almost
completely transformed from R-phase NiTi to the parent
B2 structure by 50 �C. This change was only detected by
SAD. The reverse transformation was not apparent in the
TEM image. When the foil was cooled back to room
temperature, the matrix in the corridor between the carbide
particles returned to the R-phase. At -50 �C, no B190 was
detected. After cooling to -104 �C, multivariant B190
appeared on the diffraction pattern.
TiC0.7–80NiTi
Unlike the other composites, at room temperature the
matrix of the 80NiTi composite only consisted of the R-
phase structure, whose morphology could be clearly
Fig. 31 Lens shaped B190 martensite formed after heating the TEM
foil to 108 �C and then cooling to -104 �C in TiC0.7–60NiTi
Fig. 32 Top: Matrix of 80NiTi at room temperature. Matrix has
R-phase structure. Variant structure formed in bulk condition. Middle:
T = 108 �C Matrix has B2 structure. Bottom: T = -3 �C Matrix has
R-phase structure. Variant arrangement same as in top picture, but
formed in thin-foil
5922 J Mater Sci (2008) 43:5905–5923
123
imaged. As shown in the top photo of Fig. 32, the R-phase
variants formed plates that extended away from the central
carbide particle. The accompanying SAD pattern shows
that no B190 reflections were evident.
After heating to 110 �C the R-phase reverted to the
parent B2 structure. The region shown in the middle photo
of Fig. 32 consisted of a single NiTi grain. As with all the
composites, the NiTi grain size was much larger than the
TiC0.7 particle size.
During cooling, the B2 ? R transformation started at
15 �C. By -3 �C most of the thicker areas of the matrix had
the R-phase structure. Surprisingly, the variant arrangement
after the in-situ transformation appeared to be the same as
that formed in the bulk condition. There was no apparent
thin foil effect, as had been observed in the other compos-
ites during the R ? B190 transformation. In an attempt to
observe the R ? B190 transformation in the 80NiTi sam-
ple, the temperature in the TEM was reduced to -138 �C.
No B190 reflections appeared.
Conclusions
Electron microscopy showed the details of transformation
sequence, phases, and interfaces in TiC0.7–NiTi composites
made by combustion synthesis followed by quasi-isostatic
pressing (QIP). The following conclusions can be made:
1. The interface between the TiC0.7 spheroidal phase and
the continuous NiTi phase is exceptionally clean. This
is consistent with the fracture pattern observed in the
specimens subjected to mechanical testing: cracks
propagated preferentially through the TiC0.7 particles
and not along interface with NiTi.
2. At room temperature, the matrix of the as-synthesized
20NiTi, 40NiTi, and 60NiTi composites contains both
R and B190 martensite. No distinct R-phase morphol-
ogy could be imaged.
3. In the B190 martensite, [011] Type II twinning,
(11�1ÞType I twinning and (001) compound twinning
modes were observed as the LIS of the R-B190
transformation. The [011] Type II twinning is often
reported as the LIS of the B2-B190 transformation, but
this is the first experimental confirmation of its
predicted presence as a qualified LIS of the R-B190
transformation. The (001) compound twinning mode
was the fine structure of the martensite with a wavy
morphology. Twinning was also observed at the
interface with carbide particles, which confirms that
some stress relaxation of the elastic mismatch occurs.
4. At room temperature, the matrix of the 80NiTi
composite had the R-phase structure, which appeared
with a needle-like morphology.
5. Thermal cycling resulted in the suppression of the R-
phase transformation in the TiC0.7–40NiTi composite.
This is the opposite of the behavior observed in
unreinforced NiTi alloys.
Acknowledgement This research was supported by the Director,
Office of Science, Office of Basic Energy Sciences of the US
Department of Energy under contract No. DE-AC03-76SF00098.
References
1. Merzhanov AG, Borovinskaya IP (1972) Dokl Akad Nauk SSSR
204:336
2. Gordopolov Y, Merzhanov A (1993) AIAA 154:539
3. Shikhverdiev RM, Yu A (1992) Intl J SHS 1:64
4. Munir ZA, Anselmi-Tamburini U (1989) Mater Sci Rep 3:277
5. Yi HC, Moore JJ (1990) Mater Sci 25:1159
6. Strutt ER, Olevsky EA, Meyers MA (2008) Paper 1: Combustion
synthesis/quasi-isostatic pressing of TiC–NiTi cermets: process-
ing and shrinkage anisotropy (in press)
7. Goo E, Sinclair R (1985) Acta Metall 33:1717
8. Hara T, Ohba T, Okunishi E, Otsuka K (1997) Mater Trans JIM
38:11
9. Hehemann RF, Sandrock GD (1971) Scripta Metall 5:801
10. Otsuka K, Sawamura T, Shimizu K (1971) Phys Status Solida A
5:457
11. Jacobs K, Sehitoglu H, Balszer M (1996) Metall Mater Trans A
27A:3066
12. Miyazaki S, Otsuka K, Wayman CM (1989) Acta Metall 37:1837
13. Hwang CM, Meichle M, Salamon MB, Wayman CM (1983)
Philos Mag A 47:9
14. Knowles KM, Smith DA (1981) Acta Metall 29:101
15. Onda T, Bando Y, Ohba T, Otsuka K (1992) Mater Trans JIM
33:354
16. Nishida M, Ohgi H, Itai I, Chiba A, Yamauchi K (1995) Acta
Metall Mater 43:1219
17. Xie Z, Liu Y, Van Humbeeck J (1998) Acta Mater 46:1989
18. Madangopal K, Singh JB (2000) Acta Mater 48:1325
19. Zijlstra SR, Beijer J, Klostermann JA (1974) J Mater Sci 9:145
20. Mari D, Dunand DC (1995) Metall Mater Trans A 26A:2833
21. Stroz D (1997) J Phys IV C5:C5–293
22. Miyazaki S, Igo Y, Otsuka K (1986) Acta Metall 34(10):2045
23. Gyobu A, Kawamura Y, Harikawa H, Saburi T (1996) Mater
Trans JIM 37:697
24. Honma T (1987) In: Funakubo H (ed) Shape memory alloys.
Gordon and Breach, New York, NY
25. Hanlon JE, Butler SR, Wasilewski RJ (1967) Trans Met Soc
AIME 239:1323
26. Todoroki T (1990) In: Duerig TW (ed) Engineering aspects of
shape memory alloys, Butterworth and Co., London, p 315
J Mater Sci (2008) 43:5905–5923 5923
123